PX 7.5.2 - Unable to determine CWD error

[bcarlson]

We have been getting a new error recently since we upgraded to PX 7.5.2:

Code: Select all

##F TUXX 000002 11:53:36(000) <tsort,15> Internal Error: (0):util/pathutils.C: 373: Unable to determine current working directory: Error 1728394240 occurred.
Traceback: msgAssertion__13APT_FatalPathFPCcRC11APT_UStringPCci() at 0xd376aa00
APT_getUsefulWorkingDirectory__Fb() at 0xd37dd650
initializeSystemWorkingDirectory__Fv() at 0xd3079be4
__ct__15APT_OperatorRepFP12APT_Operator() at 0xd30791ac
__ct__12APT_OperatorFv() at 0xd30adca4
__ct__22APT_CombinableOperatorFv() at 0xd31683b8
__ct__17APT_TSortOperatorFv() at 0xd2cf5d04
make__17APT_TSortOperatorFv() at 0xd2cf5ae8
instantiate__24APT_InstantiatorRegistryCFPCc() at 0xd37e960c
instantiate__22APT_PersistentRegistryFPCc() at 0xd3766840

We never used to get this until the upgrade. We have IBM AIX 5.3 64-bit and DB2 UDB 8.2.2 64 bit. DB2 runs on a separate complex of servers (currently 3 nodes, moving to 5 shortly) with a single local node on the ETL server (software versions are the same on all nodes). We have NFS mounts between each of the servers.

The problem seems to be intermittent. Sometimes restarting the failed job once works fine. Sometimes it takes a few tries. In all situations, nothing else is done other than restarting the job. That leads me to believe that it is something environmental, but I don't know where to start looking.

There are also only a handful of jobs that get this error. We probably run anywhere between 500 and 2500 DataStage jobs each day, depending on the time of the month and only a handful ever get this problem.

This happened again this morning and it was at a time when there wasn't much running on our system, so I am not sure that system load is a factor (but I won't rule it out).

Has anyone seen this before? Any suggestions?

Brad.

ps. I found one related topic, PX configuration Error, but I am not sure we have the same issue. This particular posting seemed to be related to directory permissions. Directory permissions would produce consistent errors, not intermittent. If the permissions are wrong, they are always wrong until fixed.

[ray.wurlod]

The first error message seems to suggest that PWD is not set, or its value has been "lost" somehow. I suspect you've uncovered a bug, since it only occurs intermittently.

[bcarlson]

Just a quick update. I did some more checking with our Infrastructure group. Turns out the error started showing up when we upgraded AIX to 5.2, not when we upgraded DataStage. In that case, the error may be related to something with AIX or NFS (all of our nodes are 5.2)

We are schedule to move to AIX 5.3 in the next week or two, so we'll have to see if the issue continues.

Anyway, if that sheds any new light on the subject for anyone, let me know. Thanks!

Brad.

[pneumalin]

Brad,
Did the problem get fixed after you move to AIX 5.3? Appreciated if you can let me know since we going to upgrade to 7.5.2 on AIX 5.3 very soon,
Thanks!

[thepakks]

Hi Bcarlson,

Have your error message resolved?

Kindly let's know how to figurout this.

I m also facing this problem.

What is current working directory ?
How to set this ?